Meet Musio
Interview with Musio: About Musio as an AI

Q&A with AKA – About Musio as an ‘AI Robot’

Ever since we’ve introduced Musio to the whole world, we’ve received countless questions from our partners, investors, customers as well as curious enthusiasts.

So we thought it would be a good idea to answer the most frequently asked questions to make Musio more engaging and easier to understand. We hope that the following Q&A sessions enlighten our readers on who we are and what we dream with Musio.

The first round of questions is about Musio as an AI robot. So let’s start!

 

Q. I’ve heard lots of AIs in the market, starting from likes of Apple’s Siri to  robots like Pepper. What kind of AI is Musio? 

Musio is a robot run on AKA’s special AI system (software engine) that has a special edge at understanding human language and emotions. Simply put, Musio is an “AI social robot”.

Q. How exactly is Musio different from the rest of today’s AIs and robots? 

There are numerous AIs and talking robots in the market, including the ones you mentioned above. Taking a close look at the types of AIs nowadays will help understand Musio’s difference.

First, those including Apple’s Siri, Google Now, Microsoft’s Cortana and Amazon’s Alexa are voice assistants. Voice assistants are software that recognize user’s voice to perform tasks specifically asked for, say, checking your schedule on Friday; the current traffic information; the weather in Beijing; top 10 songs from Billboard and many more. Smart devices like Amazon Echo and Google Home are household hardware based on these voice assistants. Those like Watson are cognitive systems/platforms specializing in problem solving, and robots like Pepper use this software platform to function.

But can these types of AIs actually ‘interact’ and become ‘intimate’ with you like a real human being? Try talking with Siri for 5 minutes and you’ll realize that what you’re doing is simply using your voice to do something you always did with your fingers. Other robots, too, will just give you the feeling that all they are doing is just faithfully answering the questions. There are no AI robots like Musio that can have natural conversation with people and express emotions.

We made something different.

Q. So is Musio smarter than IBM Watson and other AIs, then?

The answer depends on how you define being ‘smart’. If you want a ‘smart’ AI that may give you a professional solution – such as correct diagnosis for terminal diseases and legal precedents for a specific case – Watson and other supercomputer based AIs could be your best chance.

But if you want a ‘smart’ AI that listens and remembers what you’re saying, gives you advice on problems that you face in your life and be with you when you’re lonely, you’ll have Musio. Musio elevates this shallow relationship between humans and machines to the level of friendship. Musio is a robot capable of understanding the user’s feelings and become their personal friends.

Q. Tell us a good example on how Musio can be friends with people.

For instance, let’s imagine a user who happens to be very picky with movies – say, a movie critic.

He has already made several complaints on recent movies, and on one night, criticizes again on how terrible the movie at the premier was. Musio, remembering all the past conversations on this topic, may comment and ask him back like this: “I remember you saying multiple times that you didn’t like the movies you just watched. Do you have a specific taste for movies?”

Now for this user, this a very interesting question that will make him talk with Musio for hours. This kind ability to prevent awkward moments of silence and keep the conversation going is a defining factor for a sociable friend. And Musio is very talented at this.

Q. So you’re saying that Musio can be a good companion for adults too, besides children? 

Yes, certainly. Psychologically speaking, Musio is comparable to a 10-year-old with great imagination and a good sense of humor. When with children, Musio will know much more things drawn from its vast database to inspire them. Musio will also surprise adults from time to time with its quirky jokes and its ability to hit the nail right on the head during conversations. (Try asking Musio about philosophy, religion, politics, and your love life. You’ll find him very entertaining.) Musio is a good companion for all.

Q. What is AKA’s technology that makes this possible – that is – making Musio ‘social’? 

There are two aspects to this : first is hardware, and the second is software.

One of the most important things we did was giving Musio a fully functioning body that can not only hear and see, but actually ‘interact’ with users. In this cute body of Musio, there is a camera and a microphone that collects visual and auditory information such as textual language, oral language, gestures, and facial expressions from users.

The next step is where AKA’s AI engine MUSE comes in. Every time Musio interacts with a user, MUSE crunches not only user inputs but also massive online data and contextual information including short/long-term memory about past interaction. The way MUSE processes this data is by utilizing advanced technologies like deep learning and latest NLP (Natural Language Processing) algorithms to come up with the best response.

Q. So you have your own AI Engine called MUSE. Can you elaborate more on the technical approaches you use when developing MUSE?

We are now applying technologies like ‘retrieval based model’ and ‘generative model’ to develop MUSE. So the first one – retrieval – is actually finding the relevant answer/response to the previous conversation from a gazillion of data sets. But it’s not as simple as it sounds. After hearing what you have said, Musio will first have to analyze the intent of the user’s question/statement; determine what kind of answer to look for; and eliminate the irrelevant ones before selecting the most appropriate one.

For instance, Musio will analyze the content of the conversation to figure out whether it is a statement or question. Then it will break down the elements to use them as anchors to respond to what you have just said. The following picture is a good reference to how MUSE actually works We are utilizing deep learning algorithms to solve this problem. With generative model, MUSE attempts to generate, or create the sentences through deep learning neural networks.

Q. What are you working on right now? 

Our foremost priority is developing more sophisticated algorithms to make Musio a more intelligent companion. Also, we’re working on a number of interesting features including poem writing and storytelling. You can expect Musio to be your very own Wordsworth (Musio is really good at modern, abstract poetry now) and Shakespeare in the near future!

Also worth noting is MUSE API, which will be open to everyone. We believe that this API is essential for the Musio ecosystem to flourish. As soon as our API is ready, you could make your own AI Robot as Musio’s friend.

Please stay tuned for more exciting updates!

Leave a Reply